File size: 3,475 Bytes
8d37eb3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b2766e7
 
 
 
 
 
 
 
45ccb5b
b2766e7
 
 
45ccb5b
b2766e7
8d37eb3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
from openai import OpenAI
import streamlit as st
from utils import im_2_b64

RANDOM_SEED = 42


st.title("ChatGPT with Vision")

client = OpenAI(api_key=st.secrets["OPENAI_KEY"])

if "messages" not in st.session_state:
    st.session_state.messages = []
if "uploader_key" not in st.session_state:
    st.session_state["uploader_key"] = 0

def clear_uploader():
    st.session_state["uploader_key"] += 1
    st.rerun()

with st.sidebar:
    if st.button("Clear chat"):
        st.session_state.messages = []
        clear_uploader()

    with st.expander("Advanced Configuration"):
        st.subheader("Temperature")
        temperature = st.slider(label="x", min_value=0.1, max_value=1.0, value=0.5, step=0.1, label_visibility='hidden')
        st.subheader("Max Tokens")
        max_tokens = st.slider(label="x", min_value=32, max_value=1024, value=256, step=32, label_visibility='hidden')

    with st.expander("Image Input", expanded=True):
        images = st.file_uploader(
            "Image Upload",
            accept_multiple_files=True,
            type=["png", "jpg", "jpeg"],
            key=st.session_state["uploader_key"],
            label_visibility="collapsed",
        )


for message in st.session_state.messages:
    with st.chat_message(message["role"]):
        contents = message["content"]
        
        for content in contents:
            if content["type"] == "text":
                st.markdown(content["text"])

        number_of_images = sum(1 for c in contents if c["type"] == "image_url")
        if number_of_images > 0:
            cols = st.columns(number_of_images)
            i = 0
            for content in contents:
                if content["type"] == "image_url":
                    with cols[i]:
                        st.image(content["image_url"]["url"])
                        i += 1


def push_message(role, content, images=None):
    contents = []
    contents.append({"type": "text", "text": content})
    if images:
        for image in images:
            image_b64 = im_2_b64(image)
            image_url = f"data:image/jpeg;base64,{image_b64.decode('utf-8')}"
            obj = {
                "type": "image_url",
                "image_url": {
                    "url": image_url,
                },
            }
            contents.append(obj)

    message = {"role": role, "content": contents}
    st.session_state.messages.append(message)
    return message

chat_input_disabled = False
if prompt := st.chat_input("Type a message", key="chat_input", disabled=chat_input_disabled):
    push_message("user", prompt, images)
    with st.chat_message("user"):
        st.markdown(prompt)
        if images:
            cols = st.columns(len(images))
            for i, image in enumerate(images):
                with cols[i]:
                    st.image(image)

    with st.chat_message("assistant"):
        messages = [
            {"role": m["role"], "content": m["content"]}
            for m in st.session_state.messages
        ]
        # print("api call", messages)
        chat_input_disabled = True
        stream = client.chat.completions.create(
            model="gpt-4-vision-preview",
            messages=messages,
            stream=True,
            seed=RANDOM_SEED,
            temperature=temperature,
            max_tokens=max_tokens,
        )
        response = st.write_stream(stream)
    push_message("assistant", response)
    chat_input_disabled = False
    clear_uploader()